Effects of alloying elements on SFE

The amount of alloying elements strongly effect the SFE of these steel grades. In this study, especially the effects of manganese and carbon.elements were investigated.

Their effects were summarized below.

Manganese is the main alloying element in TWIP-steels. It preserves the austenitic structure based on the ternary system of Fe-C-Mn system, Figure 1.9.With an increasing Mn content, γ→ε transformation temperature is decreased. The main influence of Mn in TWIP steels is to control SFE; it first decreases to a minimum value and than above 20wt.%Mn increases again. Therefore, the deformation mode changes from TRIP to TWIP. At low Mn content the TRIP effect is dominant [1].

Figure 1.9 : Fe-Mn-C ternary system at 1100°C [19].

11

The aim of high Aluminium additions to high manganese steels is to enhance deformation twinning by increasing the stacking fault energy (SFE) of the steels and to improve their corrosion properties. The stacking fault probability decreases with increasing Al content, the SFE increases linearly when Al content is lower than 6.27 wt.%. Al causes the suppression of the γ→ε phase transformation and the formation of deformation twins is favoured. Al addition is very effective for improving low temperature toughness. The ductile-to-brittle transition tends to disappear with an increase in Al content, Figure 1.10. Al-addition in TWIP steel increased ductility by prohibiting the necking [28].

Figure 1.10 : Ductile- brittle transition behaviour [28].

In contrast to aluminium, silicon decreases the amount of FCC phase and sustains the γ→ε transformation during cooling and deformation. It is found that it lowered SFE of austenite, resulting in an increase in the number of stacking faults, which are nucleation sites for the ε-martensite [20].

The solubility of carbon is high in austenite, so that carbon alloying can be used to stabilize the austenite and also to strengthen the matrix by solid solution hardening. It increases the SFE and martensitic transformation is suppressed [20].

In high Mn TWIP steel, the strong attractive interaction between C and Mn leads to a non-random distribution of interstitial C atoms in which there is a very high probability that a C atom occupies an octahedral interstitial position and forms an octahedral cluster for which the number of Mn atoms on the six nearest-neighbor positions is higher than the site occupancy expected on basis of the atomic Mn

12

concentration.This clustering leads to a higher lattice resistance to dislocation glide, as the passage of a partial dislocation will in general change the local position of both substitutional and interstitial atoms, C atoms are transferred to tetrahedral sites, Figure 1.11 [30].

Figure 1.11: Illustration of reduction of the short range order (SRO) during the passage of partial dislocation on their glide plan: the octahedral clusters are shared and the C atoms are transferred to tetrahedral interstitial sites. Left: before slip, Right: after slip [14].

During the deformation of low SFE alloys, the passage of partial dislocation causes a disordering of the C-Mn clusters. Increasing planar glide results in a higher strain hardening. In addition, the process of twinning in TWIP steel interacts with the ordering, as the homogeneous twinning shear also alters the degree of short range order [30]. High manganese steels with 25wt.%Mn content have 108 different octaeder vacancies for carbon to settle and 10 different octaeder vacancies for carbon with manganese and iron atoms around. The possible coordinations of Mn and Fe atoms in octaeder are shown below, in Figure 1.12. The more Mn atoms around carbon, the stabiler is the bonding and the lower system energy. Mn-C interaction is stronger than Fe-C interaction. This bonding is antiferromagnetic below Neel-temperature [6].

Figure 1.12 : Coordinations of Fe (red) and Mn (pink) atoms in octaeder [5, 6].

13 1.2.4. Effect of temperature on SFE

Besides alloying elements, temperature has also an important effect on SFE; it increased due to the decrease in the size of the dislocation cluster when the temperature was increased to 300 K. It was also found that this behavior was reversible below 300K, however by increasing temperature (up to  390K) it was getting irreversible due to the pinning of dislocations by the carbon atoms [38].

Figure 1.13 shows the isocarbon and isomanganese diagrams of 0,6 wt.% carbon and 22 wt.% manganese.

Figure 1.13 : a) Isocarbon (6 wt pct) b) isomanganese (22 wt pct) SFE diagrams by increasing temperature. Movements of isocarbon and isomanganese lines at 300 K are seen in figures by changing interfacial energy [38].

While the isocarbon diagram in Figure 1.13(a) showes a polynomial trend of SFE against increases in the manganese content, the isomanganese diagram in Figure 1.13 b) has an almost linear increasing behavior due to increases in the carbon content.

Moreover, the effect of temperature on the isocarbon lines is more important for lower manganese contents. It can also be seen from the width of the gaps between the curves, that is, the change in SFE due to increases in temperature, which is higher for lower manganese contents [38].

Figure 1.14 shows the composition-dependent SFE maps in the Fe-Mn-C system for increases in temperature from 373 to 673 K. In a) the lines of the calculated interfacial energy could be seen. According to the equation (II) decreasing interfacial energy increases the SFE. SFE effect observed in a) at 300K completely destroys in d) at 700 K and dislocation glide is activated. As can be seen in a) with SFE value of 15-20 mJ/m² TRIP effect is activated, when higher SFE values activate TWIP effect [38].

14

Figure 1.14 : Variations in composition-dependent SFE maps by increasing temperature: (a) 373 K, (b) 473 K, (c) 573 K, and (d) 673 [38].

1.2.5 Deformation mechanisms of steels

Stacking fault energy (SFE) is an important parameter influencing deformation mechanism type. High manganese steels are plastically deformed through strain-induced martensite formation,mechanical twinning, and pure dislocation glide, due to increases in the SFE value, Figure 1.15.

If SFE is very low (≤20 mJ/m²) martensitic induced plasticity is favoured. However, higher SFE (20 mJ/m²) suppresses martensitic phase transformation and favours mechanical twinning until 60 mJ/m². And at high SFE (≥60mJ/m²) the glide of perfect dislocations is dominant because the partition of dislocations into Shockley partial dislocations is difficult [1].

Stacking fault energy of the TRIP steel is associated with ε-martensite formation, which act as nucleus of α‟-martensite, during deformation of unstable (low SFE) austenite. On the other hand, TWIP arises from occurrence of mechanical twinning during deformation of metastable (medium SFE) austenite. Therefore, it is important to choose the suitable alloying elements to obtain metastable austenite.

15

Figure 1.15 : Deformation mechanisms related to the SFE value.

1.2.5.1 Strain induced martensite formation

The concept of strain induced martensitic transformation means that martensite can form even though the temperature is above Ms. The deformation induced transformation is separated in two types; one is the stress assisted transformation where stresses help to initiate transformation, even though the temperature is above Ms and the other type is the strain induced martensitic transformation, where dislocation formation works as easier nucleation sites for the martensite [22]. The required SFE value is lower than 20mJ/m² to active the strain induced martensite formation.

The austenite phase can transform to two different types of martensite; hexagonal close packed (HCP) martensite, called ε-martensite and body centered cubic (BCC) martensite, called α„-martensite. The α„-martensite nucleate at dislocation pileups, while the nucleation of ε-martensite occurs at stacking faults. It has been known that the initial α„-martensite nuclei are coherent with the parent austenite. The ε- martensite is believed to remain coherent with the parent austenite. With enough driving force (decrease in Gibbs free energy in Eq.(I), G=200mJ/mol) α„-martensite nucleus will rapidly grow as plates. The growth of the martensite will stop at high angle grain boundaries or at other martensite plates, see Figure 1.16. This implies that the austenitic grain size is affecting the growth of martensite plates. Smaller austenitic grain size will induce smaller martensite plates. In this study, no α‟-martensite transformation was observed. The analyses showed the presence of only ε-martensite.

16

Figure 1.16 : Schematic of the growth of `-martensite plates. Their growth is stopped by other martensite plates or high angle grain boundaries

[22].

1.2.5.2 Mechanical twinning

Twinning is an important deformation mechanism inface-centered cubic (fcc) metals and alloys with a stacking fault energy (SFE) of 20-60 mJ/m². It is accepted that twinning in metals and alloys is initiated by pre-existing dislocation configurations which dissociate into multi-layered stacking fault structures forming a twin nucleus.

Dislocation-based mechanisms involve glide of Shockley partial dislocations (twinning dislocations with Burgers vector lbtwinl = ao<112>/6, where a0 is the fcc lattice constant) on successive {111} planes to create multi-layered faults that could then produce a twin. Twinning in fcc crystal occurs by shear on the {111} plane in the [112] direction. However, slip can occur on a {111} plane in either the [110] or the [1-10] direction. Figure 1.17 shows the change of stacking in the fcc lattice due to passage of partial dislocations with Burgers vector bp = btwin, during twin nucleation. Passage of the partial dislocation on the first layer creates an intrinsic stacking fault (isf), and the second and third partial dislocations on successive planes then create a three-layer twin stacking fault (tsf). Passage of more partials on neighboring {111} planes results in formation of larger twins [25].

Twinning occurs at low temperatures, in a coarse grained microstructure and at high strain rates, in presence of metastable austenite. Twins nucleate at stacking faults and materials with a low stacking fault energy have a higher tendency to form twins.

Twin formation causes grain refinement and it is an important deformation mechanism when there are not sufficient slip systems to accomodate a deformation.

17

By twinning manganese-alloyed steels can accommodate very fast at high strains.

without breaking.

Figure 1.17 : (from left to right) perfect fcc, one-layer (intrinsic) fault, two-layer fault and three-layer (twin) fault. The arrows indicate the successive {111} planes on which Shockley partials with Burgers vector bp = a0<112>/6 are passed. The translation vector T3 maintains fcc stacking between adjacent supercells [25].

There are two types of twins that can be observed in high manganese steels;

annealing twins and deformation twins, Figure 1.18 [5]. Deformation twins can be observed only after applying deformation. Crystallographically, annealing and deformation twins are identical, and consist of stacking faults on {111} planes. These faults are produced by the glide of Shockley partials [33].

Figure 1.18 : Morphologies of twinned grains: a) annealing twins; b) deformation twin [5].

1.2.5.3 Pure dislocation glide

In fcc materials primary slip occurs on {111}-planes in <110>-type directions when SFE is higher than 60 mJ/m². In fcc crystals a perfect dislocation {111} <110> can

18

dissociate in two imperfect or partial dislocations to lower its total energy and creates an incoherent lattice transition, a stacking fault, Figure 1.19. These dislocations are the result of sliding atoms moving to a location that is not a lattice point. So, the transition along the dislocation line, between the original lattice and the sheared lattice part, goes in two steps along two different <112> type directions resulting in

<110>-type dislocation after the second step. These partial dislocations are known as Shockley partial dislocation (<112>/6). In Figure 1.20 the vectors from B→C and C→B are b2 = [2-1-1] /6 and b3 = [11-2] /6 respectively. It is advantageous for a perfect dislocation with Burger‟s vector b1 = [10-1] /2 to split into two partial dislocations with Burger‟s vectors b2 and b3. The reaction b1 → b2 + b3 is indeed energetically favourable [28].

Figure 1.19 : The glide plane and stacking fault between two partial dislocations [48].

At high SFE values, the partition of dislocations into Shockley partial dislocation is difficult. Therefore, the glide of perfect dislocations is the dominant deformation mechanism. At this situation, elongation is low as a result of suppressed deformation twinning and active dislocation glide.

19

Figure 1.20 : Schematic illustration of slip on a {111} plane, [101] direction in a fcc crystal [28].

1.2.5.4 Dynamic strain aging (DSA)

The flow curve of high Mn TWIP steels containing interstitial carbon are often characterized by serrations, the Portevin–LeChatelier (PLC) phenomenon, and a negative strain rate sensitivity (SRS). Both result from a microscopic Dynamic Strain Aging (DSA) process, Figure 1.21. DSA results in an increase in flow stress and strain hardening, but a decrease in post-uniform elongation and a reduction of area at fracture. When both DSA and NSRS occur simultaneously, Portevin–

LeChatelier (PLC) bands will be observed, and the stress–strain curve will have characteristic serrations.

Figure 1.21: The interactions between DSA-PLC-SRS.

In FeMnC alloys the SRO (short range order) or SRC (short range clustering) are related to the formation of octahedral clusters by substitutional Mn (r = 140 pm) and interstitial C (r = 70 pm) atoms, The microscopic origin of the PLC effect is the dynamic strain aging (DSA) of the material due to the interaction between mobile dislocations and diffusing solute atoms. At the macroscopic scale, this dynamic strain aging leads to a negative strain rate sensitivity (SRS) of the flow stress and makes the plastic deformation nonuniform [14]. Figure 1.22 showed the Dastur-Leslie model.

20

Figure 1.22 : Dastur-Leslie model: dynamic strain aging in FeMnC alloy is caused by the reorientation of Mn-C cluster in the dislocation stress field during dislocation motion [14].

Dynamic Strain Aging (DSA) is explained as a dynamic interaction between gliding dislocations and mobile solute atoms. The serrated flow occurs during deformation in the dynamic strain aging region results from the interaction between mobile dislocations and solute atoms. The mobile dislocations that are generated during deformation are pinned by the solute atoms, and new mobile dislocations must be generated for deformation to continue [45]. This results in a much larger and more uniform dislocation density in the deformed material than that from straining at room temperature. It occurs when an aging process, related to solute atoms, is fast enough to occur during deformation, for example when the dislocation velocity is similar to the solute mobility, as dislocations move by means of thermally activated jumps, with a characteristic waiting time between two jumps. No long range diffusion is required and the solutes involved in the aging process may actually remain immobile in the lattice. If the dislocation velocity is high enough to avoid pinning by solute atoms, no DSA occurs [4].

DSA is usually not observed immediately after the start of straining, and a critical strain εC must be first achieved before the serrated flow is observed. This believed to be due to the fact that sufficient vacancies must be first created to enhance the solute diffusion to dislocations.

The Portevin-Le Chatelier (PLC) effect, also known as disordered flow, denotes a plastic instability, which is related to the discontinuous plastic flow and plastic strain inhomogenities, Figure 1.23. It can be observed in many dilute metallic solid solutions that include both interstitial and substitutional atoms. The PLC effect is usually undesirable due to its detrimental influences like loss of ductility and the appearance of surface markings on the specimen.

21

It is known that the possibility of the PLC effect reduces by decreasing the solute concentration or influencing the internal structure by precipitation. The interactions between precipitates and dislocations are assumed to cause the PLC effect. In contrast to the classical model of solute-dislocation interactions, the serrated flow is also believed to be caused directly by the shearing of precipitate, i.e., by the interactions of precipitates with mobile dislocations [43]. The microscopic origin of the PLC effect is still a matter of debate. It is mostly explained as the dynamic interaction between the moving dislocation and the diffusing solute atoms. The mobile dislocations which are carrier of the strain move disorderly between the obstacles provided by the other defects [40].

Figure 1.23 : Stress-strain curve for Al-2,5wt.%Mg alloy deformed at the strain rate 3.85×10^-4sec^-1 [9].

The deformation band formed at the center of the specimen. The aging of dislocations in the band occurred until the stress was raised to its normal level. After the stress reached a normal level, the band started to propagate through the specimen and the strain increased. The moving PLC deformation bands increased the temperature of the specimen locally due to this highly concentrated strain. This made it possible to record clear IR thermographs of the individual propagating bands which could be also observed in this study. The initiation and propagation of localized deformation bands produced a serrated flow curve [27].

In the PLC regime plastic deformation is mainly inhomogeneous and concentrated in bands. Instead of a single dislocation, a bunch of dislocations move coherently. At different strain rate and temperature, three types of bandsare observed : Type A, Type

Dislocation move disorderly between the obstacles.

22

B, Type C. When the strain rate increases and temperature decreases, types of bands change like Type C→ Type B → Type A . They are explained below, in Figure 1.24.

When the PLC effect arises, notably in some alloys and at high deformation, intermittent yielding points, or plastic instabilities, are observed, and the deformation is spatially inhomogeneous [12].Scavina et al. [41] suggested that manganese is the main alloying element and stabilizes the austenitic matrix at room temperature when carbon is employed to increment the yield and tensile strength by interstitial solute hardening.

Figure 1.24 : Characteristics of Type A, Type B, Type C Bands [9, 10].

23

In Figure 1.25 an example was given tothe PLC effect. TWIP steel showed PLC effect during uniaxial tensile tests for strain rates lower than 3x10^-1 s^-1. These plastic instabilities are probably caused by the observed negative strain-rate sensitivity (decreasing the flow stress with increasing strain rate). The plastic instabilities appear as easily visible successive surface bands; the first bands travel throughout the specimen [41].

Figure 1.25 : True stress-true strain curves of 0.5 wt pct C, 22 wt pct steel [41].

In Figure 1.26 schematic diagram of a band formation on true stress-true strain curve was given; band nucleation occurs at serrations or steps and propagates between the steps [21]. The material locally hardens during band propagation.As known, by lower strain rates, the mobile dislocations that were waiting at an obstacle - grain boundaries and twins- pinned by diffusing carbon atoms. When the applied strain rate increased, carbon atoms couldn`t diffuse fast enough to stop the dislocation mobility and the stress needed for dislocation mobility reduced, leading to negative strain rate sensitivity.

These phenomenons that resulted from a DSA, increased the strain hardening. Kim et al. [26] suggested that DSA wasn`t usually observed immediately after the start of straining and a critical strain, εC must be reached before the formation of serrated flow. This was believed to be reason of vacancy creation to provide the carbon atoms diffusion to dislocations.

24

Figure 1.26 : Schematic of a true stress-true strain curve showing that band nucleation occurs at serrations or steps. Band propagation occurs between the steps, as indicated. The material locally hardens during band propagation [21].

Ernestova [15] investigated the PLC behavior of low-alloyed steel. The investigated alloy has a chemical composition of 0,15wt.%C and 0,4wt.%Mn. As can be seen in Figure 1.27 a), by 2,5x10^-6 s^-1, the curves doesn`t show any serrations at room temperature and b) shows that marked serrations occurs at strain rates <10^-4s^-1.

In this study, tensile tests were conducted at room temperature with strain rates of 3x10^-3 s^-1, 3x10^-2 s^-1, 10^-1 s^-1, respectively and the serrations at lower strain rates could be observed more clearly.

a) b)

Figure 1.27 : Stress-strain curves on the DSA of 0,14 C wt 0,45 Mn wt a) effect of temperature at a strain rate of 2,5x10^-6 s^-1 b) effect of strain rate at a temperature of 290°C.

It has been known that the Portevin-Le Chatelier (PLC) effect is the mechanism responsible for the negative strain rate sensitivity. The occurrence of the PLC effect is clearly identified by the jerky nature of the material‟s stress-strain curve, and the macroscopic banding that develops on the material‟s surface. When a mobile dislocation is waiting at an obstacle, it may be pinned by diffusing solute atoms. An

25

increased force is required to break the dislocation free of the solute atoms, and the

increased force is required to break the dislocation free of the solute atoms, and the